400 ISO is the new 100 ISO

Lounge
1

I see so many potentially good images in the Play Raw category here on the forum that are compromised by the use of 100 ISO. In the days of film the difference between 100 ISO and 400 ISO film was huge in quality drop off as the ISO increased. But this is no longer the case with digital cameras and I feel we need to learn to embrace higher ISO numbers to improve the quality of our images.

The problems with 100 ISO are many. The first is the risk of underexposure which tends to generate huge noise problems and loss of details in the shadows. Increasing the ISO to ensure good exposure would result in less noise and this is what so many people struggle to appreciate.

The second risk is that a slow shutter speed may be required because of the low ISO. This can result in blur caused by camera shake and or subject movement. No program including darktable can really fix blur but if some noise is produced by using a higher ISO to freeze the action then darktable has great tools to tackle noise.

The third risk is too large an aperture may be needed to get correct exposure if the ISO is set to 100. This results in very shallow depth of field making focus more critical. The subject is at increased risk of being out of focus and again software can do very little with this problem. The irony is that full frame cameras are more prone to this problem because of the longer focal lengths used, but these same cameras should have the advantage of less noise with increased ISO. Despite this so many full frame users stick to 100 ISO and compromise their shots.

So what I am suggesting is that 400 ISO should become the new norm and we only lower the ISO when there is genuine need to and if necessary embrace even higher than 400 ISO when necessary. The digital world has conquered or at least tamed noise but has no real answers besides image stabilization to tackle blur and movement.

I welcome comments, alternative viewpoints and constructive criticism of my views expressed here. Maybe share your thoughts and experiences with ISO choices as this forum is about learning. Also have a good Easter break.

6 Likes
2

I think this depends heavily on what kind of photography you’re engaging in. When I’m shooting my main camera, a Nikon Z7ii, it is 98% of the time on a tripod in manual mode. I don’t shoot things that are generally moving quickly, I take my time to compose. I don’t think this camera has every been in any mode besides manual, and I look at the in-camera histogram before almost every shot. This camera is set for ISO 100, F5.6-F11, and whatever shutter suffices for ETTR pretty much all the time.

My walking around camera, a Ricoh GR III, is generally F8 @ ISO 100. The wide lens and minimal weight of the camera means I can hand hold it and generally come out with sharp photos.

I guess I still embrace the film idea of fine grain @ ISO 100. It isn’t an issue for me generally, and not much is gained or lost by changing the ISO. In the rare instance where the situations demands a higher ISO, I’m happy to do it, as I know the Z7 II can be pushed pretty high, I just never seem to need it.

I’ve been wanting to shoot more long exposures, which ISO 100 (or even lower) is actually better and will give me a longer shutter speed.

5 Likes
3

Spot on.

A lot of the time, high ISO is something I just have to live with. When shooting street, I use the zone-focus method much of the time; this means setting my aperture to f8, my shutter speed to 1/500s, and allowing the camera to select the ISO automatically. It’s not uncommon for the resulting image to get up around 1600; on bright sunny days, however, I can still get down to 200 if I’m lucky (when conditions start demanding above 1600, shutter speed has to give).

When shooting things like festivals and concerts, for example (i.e. situations where people expect to be photographed), I can often ditch the zone focusing and open up the aperture (either by going full manual or switching to auto focus). When shooting portraits, of course, I can take my time and do whatever’s required (including using a tripod and flash if necessary).

So, for me, it depends on the demands; I prefer to keep my ISO as low as I can, but shutter speed and aperture are often more important – a well-focused image taken at higher ISO is far more desirable than a blury mess of an image shot at a lower ISO; such is the trade off on the streets (although, having said that, there ARE exceptions – if one is talented/lucky enough to pull it off).

5 Likes
4

Aren’t many modern cameras ‘ISO invariant’? If I understand correctly, that means that raising the ISO when shooting, and shooting with a low ISO and then boosting it while post-processing are (almost) equivalent? For example:

750×538 48.3 KB

10 Likes
5

I agree, we were told, that ISO 100 was a must for good quality pics… and with theese settings I experienced blurry pics, thought my lenses weren’t good enough.
For me this changed with learning RawTherapee. In the meantime, I dont really care about ISO any more. Since i prefer to keep control over focus and DOF, the manual settings of aperture and shutter speed seem more important to me. Thats why I put the ISO to auto and if possible, i choose a combination of shutter speed and aperture which allow a low ISO. If not, RawTherapee can help denoising… this works for me :blush:

1 Like
6

An oldie but a goodie:

My camera’s almost ten years old, now, but is ISO invariant (it’s an X-T10). I would imagine the latest incarnations (the X-T30, X-T5, and such) have made significant inprovements with regards to ISO noise — maybe others could confirm?

2 Likes
7

The 24 MP sensor indeed improved high-ISO noise by about one stop in my experience. The 26 MP and 40 MP sensor did not improve dynamic range appreciably.

You can check exact numbers on photonstophotos.net (but you need to correct for pixel density: the X-T5 has about equal noise per pixel, but there’s three times more pixels, hence sqrt(3) times less noise across the image).

That said, more megapixels seem to work slightly better with noise reduction.

3 Likes
8

I think there are some general misconceptions around exposure and camera metering as a function of ISO settings in the digital world.

Let’s start with the hardest concept. There is no exposure triangle in digital with modern cameras (a few caveats for some analog sensor adjustments). Exposure is determined by total energy reaching the sensor. That energy is controlled by the amount of light (aperture) and the duration in time (shuttle speed). The “iso’” is controlled at your raw software using the exposure module. In the film days, the film iso controlled the response to the energy, so it did had an effect (triangle).

Therefore, for a digital raw photographer, maximizing the amount of energy captured from the scene while not clipping the sensor should be a priority.

But then the issue is metering (aka measurement). Camera meters try to find the settings for an optimal jpg so the camera can do a standard processing. It is not optimized for raw captures. This makes statements like: keep iso at 400 somewhat incorrect.

A scene. Assume it is Easter family indoor dinner and you want to capture some indoor pictures. Lighting it is not great and you have a kit lens with an 4.0 aperture.

Photographer A trained in film days sets the ISO to 1600 because it is darker inside. Aperture goes to 4 and the camera metering picks 1/1000 (auto mode).

Photographer B uses a lower iso of 400 because he read about iso irrelevant. At 4.0, the camera then picks 1/250.

Looking at a exposure triangle, both of these will be the same “exposure”. But to the raw data in the camera file, they are not. Photographer B will have more signal data (energy) to process (and yes some sensor Input referred read noise).

Photographer C sets the ISO to 100 for an indoor picture. If lighting was darker, C is then in trouble of too low of a shutter speed for hand holding, but let’s say he is ok. At 4.0, the camera will pick 1/60. Again, the same “exposure triangle”, but now he has even more raw data to process the image.

Then there is photographer D, at ISO 100 in manual mode (or better yet AutoISO). He will pick 5.6 to avoid 4.0 compromises of the lens (or use a better lens), he will keep the shutter at 1/60. He will end up under exposing, but knowing that it is only a stop and he can still get a good image in post.

Caveats: input referred read noise can be an issue; lens aperture, shutter can create a hard limit to your adjustments. Some cameras do have analog gain in the sensor sensitivity based on iso settings.

In summary, there is no 1 perfect answer. You need to develop the knowledge on what to use and why. The meter in your camera is an aid, but it lies. I wish camera had a raw data histogram.

Now, grab a camera, a tripod and setup an inside constant light scene and try this same approach. What are your observations?

Some material to read: ISO and Digital Cameras, ISO Myths www.Clarkvision.com

9 Likes
9

Yep, I agree.

Doing landscape, I shoot on a tripod unless forced otherwise (a tripod is definitely my preference*) so other than dealing with wind, I keep the ISO as low as possible – which on my decidedly non-pro Canon 850D is 100. It’s invariant but I still see more noise at 400 than 100. In fact, in a test I ran a couple of years ago 800 actually looked a bit better than 400 for long exposure nighttime images. But I never shoot nighttime images, so …

400 isn’t bad by any means and it can be cleaned up, but if it’s not necessary why bother? A little ‘normal’ noise isn’t objectionable but again if it’s avoidable, why have it?

* It’s a bit tangential at best to the original topic, but I like the discipline and forced “methodism” (even though I’m Baptist LOL) of using a tripod. Others obviously don’t, which is fine. :slight_smile:

1 Like
10

That was a very interesting read. I set up my camera in the house on a tripod and adjusted the shutter speed to 1/30th and aperture to F6.3. This gave a reasonable exposure at 25600 ISO. I then took a series of images at the following ISO settings: 100, 400, 1600, 6400, 25600.

The result is shown below. The 100 ISO is on the right and is very noisy and the 25600 is shown on the left and has the best look of the series taken. 25600 produced the best, but 6400 and 1600 ISO were not far behind. The 400 and 100ISO were definitely not as good.

When I shoot with this Canon R7 I prefer to use the Fv mode where I set the shutter speed and aperture as required and use auto ISO. Then I also have the AEB set to +/- 1.67 to hopefully ensure one exposure is suitable. I must pay attention to how closely I can process these images and which is best.

image
image715×1017 514 KB

Here is another shot taken with the Canon R7 camera set to Fv, Auto ISO and bracketing of +/- 1.67. For all intents and purposes the shots are identical when exposure compensation is applied in DT. The high ISO has clipping in the highlights (not shown in this screen grab) and this has produced problems.
image

That article was worth reading. Thanks @g-man

3 Likes
11

OK @g-man you win, you have convinced me that ISO is a lie when it comes to RAW files. Here are two shots taken at 100 ISO. The left side is overexposed by +1.67EV and the right side is under exposed by -1.67 EV. Except for the white water they are identical. The brighter one has highlights clipped and that is a problem for the white water. The rest of the image is identical.

image
image724×508 176 KB

3 Likes
12

If they were both taken at ISO 100, one underexposed, the other overexposed (and clipped), then the amount of light energy that reached the sensor was different; how does this imply / support ‘ISO is a lie’?

I think the idea is that if you take two photos, with the same aperture and shutter speed (-> same photographic exposure = same light energy), but different ISOs (say 100 and 1600), with an ISO-invariant camera, brightening the ISO 100 image by 4 EV in post-processing would give you the same image as the ISO 1600 shot, without risking overexposure.

I may be mistaken, but to me that is what the image pair I linked to above suggests.

3 Likes
13

As to the first, I am wondering what is meant by “exposure”?

On my camera, 400 ISO gets me 1/4 the exposure and 1/4 the signal to noise ratio.

14

In pro video cameras, it’s usual to not have an ISO figure, but the gain of the pixel amplifiers.

Increasing the gain on one of these low noise amplifiers does increase the noise (as any amplifier amplifies the noise and adds to it).

2 Likes
15

The X-T3 had a dual gain ISO, where noise was markedly reduced at ISO 800, so there were essentially two “base” ISOs at 160 and 800. ETTR dictates to use either one of these, depending on light levels.

(all from memory, I may have gotten the camera version and threshold wrong)

4 Likes
16

You’re right, it has a second base ISO at 800 for some savings in dynamic range. You can see it from the chart in Photographic Dynamic Range versus ISO Setting.

3 Likes
17

My two old Sigma cameras are what I believe is called “ISO-less” meaning in my case that the raw histogram of a properly exposed capture at 100 ISO is reasonably close to max but not clipped at the high end. At ISO 400, the raw histogram for the same scene shows the sensor to be under-exposed -2 EV while the raw conversion to RGB is brightened by 2 EV to compensate.

According to the sensor data sheet, the noise is 70e- and well capacity is 70,000e-. About a 40,00e- limit is recommended for linearity for that sensor, so DR = 9 EV max for that sensor for 100 ISO but only 7 EV if I choose 400 ISO. Ergo, I leave it at 100 ISO, only rarely selecting more.

18

I don’t understand that. I would understand no change in the histogram (the camera simply boosting brightness when generating the JPG output, ‘applying the ISO’), or I could understand if the ISO 400 histogram were blown, but underexposed by 2 EV?!

19

I said the raw histogram as shown in for example RawDigger - not what is shown on the camera LCD or on the monitor screen.

ISO says: EI = 10/Ha
where Ha is the arithmetic mean focal plane exposure, expressed in lux-seconds (lx⋅s)

Therefore, exposure Ha = 10/EI, where EI represents the ISO setting.

So, at 100 ISO. an average exposure is 0.1 lux-sec and, at 400 ISO, it is 0.025 lux-sec a ratio of -2 EV

Hope this helps you to understand what I said.

20

You’re too easily convinced. :wink: :wink:
@g-man did mention -

From your first example (and from the DPReview review), the R7 is clearly not fully ISO-invariant, so you are better off using a higher ISO where needed, as opposed to the underexpose-then-push approach. I’ve been trying to find out if maybe it has a dual-gain structure or similar that would basically mean you have 2 base ISOs, but haven’t found out yet.